K9-Overwatch

A lost-and-found pet platform for the Indianapolis metro that scrapes five separate pet-listing sources — each one reverse-engineered differently — geocodes them onto one map, and matches lost dogs against found reports.

🖥 live demopublic repo

View on GitHub ↗

The Build Loop

Because I saw how scattered and frantic lost-pet searches are, I kept hitting the fact that a lost dog might be reported on any of five different sites, none of which talk to each other, so I built a platform that aggregates all five into one geographic view and automatically matches lost against found, that gives owners one place to look instead of five. It taught me that real aggregation means solving each source on its own terms — one had an open API, one hid behind AWS WAF, one needed a headless browser.

What it looks like

Map view — lost and found pet reports plotted from multiple scraped sources.
Pet listings — normalized records aggregated from shelters and lost-pet boards.
Match candidates — the matcher pairing lost reports against found reports and flagging cross-source duplicates.
Recorded walk-through of the running app.

K9-Overwatch

One-liner: A lost-and-found pet platform for the Indianapolis metro that scrapes five separate pet-listing sources — each one reverse-engineered differently — geocodes them onto one map, and matches lost dogs against found reports.

Origin (the Build Loop)

Because I saw how scattered and frantic lost-pet searches are, I kept hitting the fact that a lost dog might be reported on any of five different sites, none of which talk to each other, so I built a platform that aggregates all five into one geographic view and automatically matches lost against found, that gives owners one place to look instead of five. It taught me that real aggregation means solving each source on its own terms — one had an open API, one hid behind AWS WAF, one needed a headless browser.

The problem

When a pet goes missing, reports land across a patchwork of services — shelters, PawBoost, local alert sites, Pet FBI — with no shared map and no cross-matching. An owner has to check all of them, repeatedly. I wanted one unified, geocoded view that also did the matching work automatically, which meant pulling from five sources that each guard their data differently.

What it does

  • Aggregates lost / found / adoptable pet listings from 5 sources into one database.
  • Geocodes street-level addresses to lat/lon coordinates.
  • Displays pets on an interactive map with filtering by type, status, color, size, and recency.
  • Matches lost pets against found/shelter records using description similarity.
  • Alerts users when a new pet matching their criteria appears near their location.

How it’s built

  • Stack: Python; per-source scrapers/clients; geocoding; interactive map front end.
  • Notable engineering — five sources, five techniques (documented in docs/):
    • 24petconnect — PetHarbor backend, HTML scraping.
    • PawBoost — Cloudflare-protected, Playwright required.
    • IndyLostPetAlert — open WordPress REST API.
    • Pet FBI — GraphQL API behind AWS WAF; provides lat/lon directly (skips geocoding).
    • Lost My Doggie — Cloudflare-protected phone-alert service.
    • Plus Haversine distance math for the “near me” matching/alerts.
  • Architecture: 5 source adapters → normalize + geocode → unified DB → similarity matcher → interactive map + proximity alerts.

Proof points

  • 5 heterogeneous sources unified — open REST, GraphQL/WAF, HTML scrape, and two Cloudflare-protected sites.
  • Description-similarity matching plus Haversine proximity alerts.
  • Live and functional (not shelved).

What to show

  • Demo: Deploy to a public URL (repo is public). The interactive map with real aggregated Indy-area listings is the hero view.
  • Visuals needed: the map with filters applied; a lost→found match example; the source-analysis table from docs/.

Case-study angle

A genuinely useful community tool that doubles as proof of integration range: five services, five different access methods — open API to AWS-WAF GraphQL to real-browser rendering of JavaScript-heavy sites — unified into one map that reunites people with their pets.